0
点赞
收藏
分享

微信扫一扫

MySQL 8.0 高可用之MGR（组复制）介绍

归零者245号 2022-01-07 阅读 48
mysql主从复制半同步复制MGR

作者公众号，欢迎一起交流。

1 背景

MySQL Group Replication（简称MGR）是MySQL官方于2016年12月12日推出的一款高可用与高扩展的解决方案，它提供了高可用、高扩展、高可靠的MySQL集群服务，具有以下特点：

  • 强一致性：基于原生复制和paxos协议，保证数据传输的一致性和原子性
  • 高容错性：只要大多数节点没有出现故障，集群就可对外提供服务
  • 高扩展性：节点的加入和移除都是自动的，不需要人为过多干预
  • 高灵活性：具有单主模式和多主模式，单主模式在主宕机后自动选主，多主模式多节点写入

2 组复制插件架构

MySQL MGR是一个MySQL插件，构建在现有MySQL复制基础架构上。MGR插件包括一组用于Capture、Apply和Lifecycle的API，用于控制插件如何与MySQL服务器进行交互。通过接口可以让信息在服务器和插件之间流转，而且这些接口将MySQL服务器核心从组复制插件隔离出来。从服务器到插件这个方向，会有如服务器启动、服务器恢复、服务器准备接受连接和服务器即将提交事务的事件的通知，而在另一个方向，即插件到服务器这个方向，插件指示服务器执行如提交或终止正在进行的事务，或在Relaylog中对事务排队这样的操作。

再往下一层是一组组件，包括Capture、Applier和Recovery，当通知路由到这些组件时会做出相应的响应。其中，Capture组件负责追踪正在执行事务的上下文信息，Applier组件负责在数据库上执行远程事务，Recover组件管理分布式恢复。

继续往下看，复制协议模块包含复制协议的特定逻辑，它处理冲突检测，接收事务并将其传播到组。

最后两层是组通信系统（GCS）API，以及基于Paxos的组通信引擎的实现。

 3 配置要求和限制

组复制的使用需要满足如下要求：

  • InnoDB存储引擎
  • 每个表必须有显式主键
  • 网络性能要求低延时、高带宽
  • server_id要唯一
  • 开启GTID
  • 事务隔离级别建议使用RC
  • 不建议使用外键级联约束
  • 组成员最大数目为9

4 单主模式部署

4.1 部署环境

Hostname

IP

Server ID

Port

node1

192.168.56.201

201

3306

node2

192.168.56.202

202

3306

node3

192.168.56.203

203

3306

4.2 安装部署

4.2.1 安装MySQL数据库软件

1）分别在节点node1、node2和node3的/etc/hosts文件增加如下配置：

192.168.56.201 node1 node1.com.cn

192.168.56.202 node2 node2.com.cn

192.168.56.203 node3 node3.com.cn

2）分别在节点node1、node2和node3关闭防火墙：

查看防火墙状态：

systemctl status firewalld

关闭防火墙：

systemctl stop firewalld

禁用防火墙启动：

systemctl disable firewalld

3）安装MySQL数据库，参照【MySQL 8.0 数据库安装部署】分别在节点node1、node2和node3安装MySQL数据库软件。

4.2.2 安装主节点node1

1）配置组复制，在原有的配置文件/etc/my.cnf中增加如下配置信息：

disabled_storage_engines = "MyISAM,BLACKHOLE,FEDERATED,ARCHIVE,MEMORY"

gtid_mode = ON

enforce_gtid_consistency = ON

binlog_checksum = CRC32

plugin_load_add = 'group_replication.so'

group_replication_group_name = "0b15d45f-6e29-11ec-99b3-08002790c7d5"

group_replication_start_on_boot = off

group_replication_local_address = "node1:33061"

group_replication_group_seeds = "node1:33061,node2:33061,node3:33061"

group_replication_bootstrap_group = off

2）重启MySQL服务器实例使配置生效：

[root@node1 ~]# service mysql.server restart

Shutting down MySQL.... SUCCESS!

Starting MySQL....... SUCCESS!

3）连接MySQL服务器创建分布式恢复用户并授予权限，若使用Clone操作，需授予对应的权限：

mysql> set sql_log_bin=0;

Query OK, 0 rows affected (0.00 sec)



mysql> create user repl identified by 'repl';

Query OK, 0 rows affected (0.12 sec)



mysql> grant replication slave on *.* to repl;

Query OK, 0 rows affected (0.11 sec)



mysql> grant backup_admin on *.* to repl;

Query OK, 0 rows affected (0.10 sec)



mysql> set sql_log_bin=1;

Query OK, 0 rows affected (0.00 sec)



mysql> change replication source to source_user='repl',source_password='repl' for channel 'group_replication_recovery';

Query OK, 0 rows affected, 2 warnings (0.17 sec)

4）引导组，第一次启动组时需使用group_replication_bootstrap_group系统变量引导组，该操作只能在单个服务器完成，并且只能启动一次，引导完成查看组成员信息：

mysql> set global group_replication_bootstrap_group=on;

Query OK, 0 rows affected (0.00 sec)



mysql> start group_replication;

Query OK, 0 rows affected (1.29 sec)



mysql> set global group_replication_bootstrap_group=off;

Query OK, 0 rows affected (0.00 sec)



mysql> select * from performance_schema.replication_group_members\G;

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: a6886158-6ef5-11ec-a75e-08002790c7d5

MEMBER_HOST: node1.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

1 row in set (0.12 sec)



ERROR:

No query specified

5）出于演示目的，创建测试表：

mysql> create database test;

Query OK, 1 row affected (0.18 sec)



mysql> use test;

Database changed

mysql> create table t1(id int primary key,name varchar(20));

Query OK, 0 rows affected (0.21 sec)



mysql> insert into t1 values(1,'Alen');

Query OK, 1 row affected (0.03 sec)



mysql> show binary logs;

+------------------+-----------+-----------+

| Log_name | File_size | Encrypted |

+------------------+-----------+-----------+

| mysql-bin.000001 | 179 | No |

| mysql-bin.000002 | 500 | No |

| mysql-bin.000003 | 1190 | No |

+------------------+-----------+-----------+

3 rows in set (0.00 sec)



mysql> show binlog events in 'mysql-bin.000003';

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

| Log_name | Pos | Event_type | Server_id | End_log_pos | Info |

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

| mysql-bin.000003 | 4 | Format_desc | 201 | 125 | Server ver: 8.0.27, Binlog ver: 4 |

| mysql-bin.000003 | 125 | Previous_gtids | 201 | 156 | |

| mysql-bin.000003 | 156 | Gtid | 201 | 242 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:1' |

| mysql-bin.000003 | 242 | Query | 201 | 308 | BEGIN |

| mysql-bin.000003 | 308 | View_change | 201 | 411 | view_id=16414770057187861:1 |

| mysql-bin.000003 | 411 | Query | 201 | 483 | COMMIT |

| mysql-bin.000003 | 483 | Gtid | 201 | 567 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:2' |

| mysql-bin.000003 | 567 | Query | 201 | 675 | create database test /* xid=25 */ |

| mysql-bin.000003 | 675 | Gtid | 201 | 759 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:3' |

| mysql-bin.000003 | 759 | Query | 201 | 899 | use `test`; create table t1(id int primary key,name varchar(20)) /* xid=30 */ |

| mysql-bin.000003 | 899 | Gtid | 201 | 985 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:4' |

| mysql-bin.000003 | 985 | Query | 201 | 1060 | BEGIN |

| mysql-bin.000003 | 1060 | Table_map | 201 | 1114 | table_id: 93 (test.t1) |

| mysql-bin.000003 | 1114 | Write_rows | 201 | 1159 | table_id: 93 flags: STMT_END_F |

| mysql-bin.000003 | 1159 | Xid | 201 | 1190 | COMMIT /* xid=31 */ |

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

15 rows in set (0.00 sec)

4.2.3 新增从节点node2

新增从节点node2的操作和node1相似，可参照node1的配置过程，只不过不再需要引导组了。

1）配置组复制，在原有节点node2的配置文件/etc/my.cnf中增加如下配置信息，需修改server_id等配置选项：

server_id = 202

disabled_storage_engines = "MyISAM,BLACKHOLE,FEDERATED,ARCHIVE,MEMORY"

gtid_mode = ON

enforce_gtid_consistency = ON

binlog_checksum = CRC32

plugin_load_add = 'group_replication.so'

group_replication_group_name = "0b15d45f-6e29-11ec-99b3-08002790c7d5"

group_replication_start_on_boot = off

group_replication_local_address = "node2:33061"

group_replication_group_seeds = "node1:33061,node2:33061,node3:33061"

group_replication_bootstrap_group = off

2）重启MySQL服务器实例使配置生效：

[root@node2 ~]# service mysql.server restart

Shutting down MySQL.. SUCCESS!

Starting MySQL...... SUCCESS!

3）连接MySQL服务器创建分布式恢复用户并授予权限，若使用Clone操作，需授予对应的权限：

mysql> set sql_log_bin=0;

Query OK, 0 rows affected (0.00 sec)



mysql> create user repl identified by 'repl';

Query OK, 0 rows affected (0.06 sec)



mysql> grant replication slave on *.* to repl;

Query OK, 0 rows affected (0.01 sec)



mysql> grant backup_admin on *.* to repl;

Query OK, 0 rows affected (0.11 sec)



mysql> set sql_log_bin=1;

Query OK, 0 rows affected (0.00 sec)



mysql> change replication source to source_user='repl',source_password='repl'\

-> for channel 'group_replication_recovery';

Query OK, 0 rows affected, 2 warnings (0.19 sec)

4）启动组复制并查看成员状态：

mysql> start group_replication;

Query OK, 0 rows affected (6.95 sec)



mysql> select * from performance_schema.replication_group_members\G;

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 4bfafef9-6efa-11ec-8296-080027e8893b

MEMBER_HOST: node2.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: SECONDARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: a6886158-6ef5-11ec-a75e-08002790c7d5

MEMBER_HOST: node1.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

2 rows in set (0.16 sec)



ERROR:

No query specified

5）验证从node1节点同步过来的数据：

mysql> select * from test.t1;

+----+------+

| id | name |

+----+------+

| 1 | Alen |

+----+------+

1 row in set (0.00 sec)



mysql> show binlog events in 'mysql-bin.000003';

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

| Log_name | Pos | Event_type | Server_id | End_log_pos | Info |

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

| mysql-bin.000003 | 4 | Format_desc | 202 | 125 | Server ver: 8.0.27, Binlog ver: 4 |

| mysql-bin.000003 | 125 | Previous_gtids | 202 | 156 | |

| mysql-bin.000003 | 156 | Gtid | 201 | 242 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:1' |

| mysql-bin.000003 | 242 | Query | 201 | 308 | BEGIN |

| mysql-bin.000003 | 308 | View_change | 201 | 415 | view_id=16414770057187861:1 |

| mysql-bin.000003 | 415 | Query | 201 | 487 | COMMIT |

| mysql-bin.000003 | 487 | Gtid | 201 | 571 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:2' |

| mysql-bin.000003 | 571 | Query | 201 | 679 | create database test /* xid=31 */ |

| mysql-bin.000003 | 679 | Gtid | 201 | 763 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:3' |

| mysql-bin.000003 | 763 | Query | 201 | 903 | use `test`; create table t1(id int primary key,name varchar(20)) /* xid=32 */ |

| mysql-bin.000003 | 903 | Gtid | 201 | 989 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:4' |

| mysql-bin.000003 | 989 | Query | 201 | 1059 | BEGIN |

| mysql-bin.000003 | 1059 | Table_map | 201 | 1113 | table_id: 90 (test.t1) |

| mysql-bin.000003 | 1113 | Write_rows | 201 | 1158 | table_id: 90 flags: STMT_END_F |

| mysql-bin.000003 | 1158 | Xid | 201 | 1189 | COMMIT /* xid=34 */ |

| mysql-bin.000003 | 1189 | Gtid | 201 | 1275 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:5' |

| mysql-bin.000003 | 1275 | Query | 201 | 1341 | BEGIN |

| mysql-bin.000003 | 1341 | View_change | 201 | 1554 | view_id=16414770057187861:2 |

| mysql-bin.000003 | 1554 | Query | 201 | 1626 | COMMIT |

| mysql-bin.000003 | 1626 | Gtid | 201 | 1712 | SET @@SESSION.GTID_NEXT= '0b15d45f-6e29-11ec-99b3-08002790c7d5:6' |

| mysql-bin.000003 | 1712 | Query | 201 | 1778 | BEGIN |

| mysql-bin.000003 | 1778 | View_change | 201 | 1925 | view_id=16414770057187861:4 |

| mysql-bin.000003 | 1925 | Query | 201 | 1997 | COMMIT |

+------------------+------+----------------+-----------+-------------+-------------------------------------------------------------------------------+

23 rows in set (0.00 sec)

4.2.4 新增从节点node3

1）新增从节点node3的操作可参考从节点node2的添加过程，仅仅是配置选项不同，其它操作过程一样，node3配置文件/etc/my.cnf配置选项如下：

server_id = 203

disabled_storage_engines = "MyISAM,BLACKHOLE,FEDERATED,ARCHIVE,MEMORY"

gtid_mode = ON

enforce_gtid_consistency = ON

binlog_checksum = CRC32

plugin_load_add = 'group_replication.so'

group_replication_group_name = "0b15d45f-6e29-11ec-99b3-08002790c7d5"

group_replication_start_on_boot = off

group_replication_local_address = "node3:33061"

group_replication_group_seeds = "node1:33061,node2:33061,node3:33061"

group_replication_bootstrap_group = off

2）新增从节点node3完成后，查看组成员信息，都处于在线状态：

mysql> select * from performance_schema.replication_group_members\G;

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 07cad00b-6efd-11ec-87cc-08002715e2e8

MEMBER_HOST: node3.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: SECONDARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 4bfafef9-6efa-11ec-8296-080027e8893b

MEMBER_HOST: node2.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: SECONDARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 3. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: a6886158-6ef5-11ec-a75e-08002790c7d5

MEMBER_HOST: node1.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

3 rows in set (0.01 sec)



ERROR:

No query specified

5 单主 / 多主模式切换

MySQL MGR默认运行在单主模式，即参数group_replication_single_primary_mode的值默认是ON，自MySQL 8.0.13开始，可以在MySQL MGR运行的情况下利用group_replication_switch_to_multi_primary_mode()和group_replication_switch_to_single_primary_mode()函数进行单主/多主的模式切换。

5.1 单主切换为多主

使用函数group_replication_switch_to_multi_primary_mode()可将单主模式切换为多主模式：

mysql> select group_replication_switch_to_multi_primary_mode();

+--------------------------------------------------+

| group_replication_switch_to_multi_primary_mode() |

+--------------------------------------------------+

| Mode switched to multi-primary successfully. |

+--------------------------------------------------+

1 row in set (1.00 sec)



mysql> select * from performance_schema.replication_group_members\G;

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 07cad00b-6efd-11ec-87cc-08002715e2e8

MEMBER_HOST: node3.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 4bfafef9-6efa-11ec-8296-080027e8893b

MEMBER_HOST: node2.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 3. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: a6886158-6ef5-11ec-a75e-08002790c7d5

MEMBER_HOST: node1.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

3 rows in set (0.00 sec)



ERROR:

No query specified

5.2 多主切换为单主

使用函数group_replication_switch_to_single_primary_mode() 可将多主模式切换为单主模式：

mysql> select group_replication_switch_to_single_primary_mode();

+---------------------------------------------------+

| group_replication_switch_to_single_primary_mode() |

+---------------------------------------------------+

| Mode switched to single-primary successfully. |

+---------------------------------------------------+

1 row in set (0.03 sec)



mysql> select * from performance_schema.replication_group_members\G;

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 07cad00b-6efd-11ec-87cc-08002715e2e8

MEMBER_HOST: node3.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: PRIMARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 4bfafef9-6efa-11ec-8296-080027e8893b

MEMBER_HOST: node2.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: SECONDARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

*************************** 3. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: a6886158-6ef5-11ec-a75e-08002790c7d5

MEMBER_HOST: node1.com.cn

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

MEMBER_ROLE: SECONDARY

MEMBER_VERSION: 8.0.27

MEMBER_COMMUNICATION_STACK: XCom

3 rows in set (0.00 sec)



ERROR:

No query specified

6 监视组复制环境

对于MySQL MGR的监视，主要是通过performance_schema下的表来实现，主要的表如下：

mysql> show tables from performance_schema like '%replication%';

+------------------------------------------------------+

| Tables_in_performance_schema (%replication%) |

+------------------------------------------------------+

| replication_applier_configuration |

| replication_applier_filters |

| replication_applier_global_filters |

| replication_applier_status |

| replication_applier_status_by_coordinator |

| replication_applier_status_by_worker |

| replication_asynchronous_connection_failover |

| replication_asynchronous_connection_failover_managed |

| replication_connection_configuration |

| replication_connection_status |

| replication_group_communication_information |

| replication_group_configuration_version |

| replication_group_member_actions |

| replication_group_member_stats |

| replication_group_members |

+------------------------------------------------------+

15 rows in set (0.01 sec)
举报

相关推荐

MySQL 8.0 MGR（组复制）高可用VIP故障转移脚本

phpMySQL配置文件mgr组复制数据库

MySQL高可用--组复制

MySQL高可用性组复制数据库

MGR Mysql 组复制

mysql数据库运维yyds干货盘点

dba+开源工具：MySQL 8.0 MGR高可用VIP切换脚本

phpmysql配置文件MySQL数据库

【高并发高性能高可用之海量数据MySQL实战-1】-MySQL 8.0安装

mysql数据库database

redis高可用之主从复制

后端中间件分布式数据redisHarmonyOS后端开发

MySQL高可用之 MMM结构

mysql数据库

MySQL高可用架构MGR+ProxySQL+Keepalived

MySQLLinux高可用

【高并发高性能高可用之海量数据MySQL实战-30】-MySQL基于主从复制的高可用方案

mysql数据库database

Redis高可用之主从复制实践

技术分享Java
0 条评论